Question 2
library(palaeoSig)
## Loading required package: vegan
## Loading required package: permute
## Loading required package: lattice
## This is vegan 2.5-2
library(vegan)
library(analogue)
## analogue version 0.17-0
library(rioja)
## This is rioja 0.9-15.1
##
## Attaching package: 'rioja'
## The following objects are masked from 'package:analogue':
##
## crossval, performance
data(arctic.pollen)
data("arctic.env")
arctic_sqrt <- sqrt(arctic.pollen)
arctic_cca_tmin <- vegan::cca(arctic_sqrt ~ arctic.env$tmin)
arctic_cca_tmin_sum <- summary(arctic_cca_tmin)
arctic_cca_tmin_sum$cont
## $importance
## Importance of components:
## CCA1 CA1 CA2 CA3 CA4 CA5
## Eigenvalue 0.03753 0.2435 0.1348 0.06629 0.04657 0.03796
## Proportion Explained 0.03290 0.2134 0.1182 0.05810 0.04081 0.03327
## Cumulative Proportion 0.03290 0.2463 0.3644 0.42255 0.46336 0.49663
## CA6 CA7 CA8 CA9 CA10 CA11
## Eigenvalue 0.03694 0.03355 0.03299 0.02860 0.02816 0.02608
## Proportion Explained 0.03238 0.02940 0.02892 0.02506 0.02468 0.02286
## Cumulative Proportion 0.52901 0.55841 0.58733 0.61240 0.63707 0.65994
## CA12 CA13 CA14 CA15 CA16 CA17
## Eigenvalue 0.02509 0.02339 0.02246 0.02088 0.02027 0.01894
## Proportion Explained 0.02199 0.02050 0.01969 0.01831 0.01777 0.01660
## Cumulative Proportion 0.68192 0.70242 0.72211 0.74041 0.75818 0.77478
## CA18 CA19 CA20 CA21 CA22 CA23
## Eigenvalue 0.01854 0.01800 0.01721 0.01609 0.01584 0.01457
## Proportion Explained 0.01625 0.01578 0.01508 0.01411 0.01388 0.01277
## Cumulative Proportion 0.79103 0.80680 0.82188 0.83599 0.84987 0.86264
## CA24 CA25 CA26 CA27 CA28 CA29
## Eigenvalue 0.01369 0.01296 0.01285 0.01252 0.01158 0.011311
## Proportion Explained 0.01200 0.01136 0.01126 0.01097 0.01015 0.009914
## Cumulative Proportion 0.87464 0.88600 0.89726 0.90824 0.91839 0.928299
## CA30 CA31 CA32 CA33 CA34
## Eigenvalue 0.010854 0.010533 0.010087 0.009837 0.009288
## Proportion Explained 0.009513 0.009232 0.008841 0.008622 0.008141
## Cumulative Proportion 0.937812 0.947044 0.955885 0.964507 0.972648
## CA35 CA36 CA37 CA38
## Eigenvalue 0.008369 0.008206 0.007512 0.007119
## Proportion Explained 0.007336 0.007193 0.006584 0.006240
## Cumulative Proportion 0.979984 0.987176 0.993760 1.000000
arctic_cca_tmax <- vegan::cca(arctic_sqrt ~ arctic.env$tmax)
arctic_cca_tmax_sum <- summary(arctic_cca_tmax)
arctic_cca_tmax_sum$cont
## $importance
## Importance of components:
## CCA1 CA1 CA2 CA3 CA4 CA5
## Eigenvalue 0.1761 0.1438 0.08789 0.07051 0.05173 0.04362
## Proportion Explained 0.1544 0.1260 0.07703 0.06180 0.04534 0.03823
## Cumulative Proportion 0.1544 0.2804 0.35741 0.41921 0.46455 0.50279
## CA6 CA7 CA8 CA9 CA10 CA11
## Eigenvalue 0.03428 0.03372 0.03234 0.02878 0.02632 0.02572
## Proportion Explained 0.03005 0.02956 0.02834 0.02523 0.02307 0.02255
## Cumulative Proportion 0.53284 0.56239 0.59074 0.61597 0.63904 0.66158
## CA12 CA13 CA14 CA15 CA16 CA17
## Eigenvalue 0.02450 0.02333 0.02155 0.02099 0.02059 0.01910
## Proportion Explained 0.02147 0.02045 0.01889 0.01840 0.01805 0.01674
## Cumulative Proportion 0.68305 0.70350 0.72239 0.74079 0.75884 0.77557
## CA18 CA19 CA20 CA21 CA22 CA23
## Eigenvalue 0.01862 0.01777 0.01720 0.01632 0.01539 0.01462
## Proportion Explained 0.01632 0.01558 0.01508 0.01430 0.01349 0.01281
## Cumulative Proportion 0.79189 0.80747 0.82254 0.83685 0.85034 0.86315
## CA24 CA25 CA26 CA27 CA28 CA29
## Eigenvalue 0.01339 0.01323 0.01285 0.01238 0.01154 0.011309
## Proportion Explained 0.01173 0.01160 0.01126 0.01085 0.01012 0.009912
## Cumulative Proportion 0.87488 0.88648 0.89774 0.90859 0.91871 0.928621
## CA30 CA31 CA32 CA33 CA34
## Eigenvalue 0.010956 0.010457 0.009957 0.009545 0.009274
## Proportion Explained 0.009603 0.009165 0.008727 0.008366 0.008128
## Cumulative Proportion 0.938224 0.947389 0.956116 0.964482 0.972610
## CA35 CA36 CA37 CA38
## Eigenvalue 0.008495 0.008258 0.007549 0.006948
## Proportion Explained 0.007446 0.007238 0.006617 0.006090
## Cumulative Proportion 0.980055 0.987293 0.993910 1.000000
arctic_cca_pann <- vegan::cca(arctic_sqrt ~ arctic.env$ptotal)
arctic_cca_pann_sum <- summary(arctic_cca_pann)
arctic_cca_pann_sum$cont
## $importance
## Importance of components:
## CCA1 CA1 CA2 CA3 CA4 CA5
## Eigenvalue 0.06729 0.2131 0.1354 0.06644 0.04646 0.04140
## Proportion Explained 0.05898 0.1868 0.1187 0.05824 0.04072 0.03629
## Cumulative Proportion 0.05898 0.2458 0.3644 0.42265 0.46337 0.49966
## CA6 CA7 CA8 CA9 CA10 CA11
## Eigenvalue 0.03546 0.03375 0.03283 0.02903 0.0275 0.02614
## Proportion Explained 0.03108 0.02958 0.02877 0.02545 0.0241 0.02291
## Cumulative Proportion 0.53074 0.56033 0.58910 0.61455 0.6386 0.66156
## CA12 CA13 CA14 CA15 CA16 CA17
## Eigenvalue 0.02483 0.02317 0.02263 0.02085 0.01966 0.01880
## Proportion Explained 0.02176 0.02031 0.01984 0.01827 0.01723 0.01648
## Cumulative Proportion 0.68332 0.70363 0.72347 0.74174 0.75897 0.77545
## CA18 CA19 CA20 CA21 CA22 CA23
## Eigenvalue 0.01819 0.01792 0.01721 0.01622 0.01566 0.01440
## Proportion Explained 0.01595 0.01571 0.01508 0.01421 0.01372 0.01262
## Cumulative Proportion 0.79139 0.80710 0.82218 0.83640 0.85012 0.86274
## CA24 CA25 CA26 CA27 CA28 CA29
## Eigenvalue 0.01372 0.01317 0.01284 0.01248 0.01178 0.011297
## Proportion Explained 0.01203 0.01155 0.01126 0.01094 0.01032 0.009902
## Cumulative Proportion 0.87477 0.88632 0.89758 0.90851 0.91884 0.928737
## CA30 CA31 CA32 CA33 CA34
## Eigenvalue 0.010712 0.010471 0.010102 0.009840 0.009272
## Proportion Explained 0.009389 0.009178 0.008854 0.008624 0.008126
## Cumulative Proportion 0.938126 0.947304 0.956158 0.964782 0.972909
## CA35 CA36 CA37 CA38
## Eigenvalue 0.008308 0.007962 0.007548 0.007091
## Proportion Explained 0.007282 0.006978 0.006616 0.006215
## Cumulative Proportion 0.980191 0.987169 0.993785 1.000000
tmax explains the most variance, because the ratio of CCA1/CA1 is the highest,
arctic_cca_tjul <- vegan::cca(arctic_sqrt ~ arctic.env$tjul)
arctic_cca_tjul_sum <- summary(arctic_cca_tjul)
arctic_cca_tjul_sum$cont
## $importance
## Importance of components:
## CCA1 CA1 CA2 CA3 CA4 CA5
## Eigenvalue 0.1530 0.1547 0.09566 0.07269 0.05484 0.04373
## Proportion Explained 0.1341 0.1356 0.08384 0.06371 0.04806 0.03832
## Cumulative Proportion 0.1341 0.2697 0.35349 0.41720 0.46527 0.50359
## CA6 CA7 CA8 CA9 CA10 CA11
## Eigenvalue 0.03706 0.03312 0.03183 0.02889 0.02617 0.02573
## Proportion Explained 0.03248 0.02903 0.02790 0.02532 0.02293 0.02256
## Cumulative Proportion 0.53607 0.56510 0.59300 0.61832 0.64126 0.66381
## CA12 CA13 CA14 CA15 CA16 CA17
## Eigenvalue 0.02362 0.02275 0.02196 0.02068 0.02044 0.01879
## Proportion Explained 0.02071 0.01994 0.01925 0.01813 0.01792 0.01647
## Cumulative Proportion 0.68452 0.70446 0.72370 0.74183 0.75975 0.77622
## CA18 CA19 CA20 CA21 CA22 CA23
## Eigenvalue 0.01862 0.01791 0.01711 0.01633 0.01488 0.01449
## Proportion Explained 0.01632 0.01570 0.01500 0.01431 0.01304 0.01270
## Cumulative Proportion 0.79254 0.80823 0.82323 0.83754 0.85058 0.86329
## CA24 CA25 CA26 CA27 CA28 CA29
## Eigenvalue 0.01352 0.01317 0.01281 0.01239 0.011328 0.011098
## Proportion Explained 0.01185 0.01154 0.01123 0.01086 0.009929 0.009727
## Cumulative Proportion 0.87514 0.88668 0.89791 0.90877 0.918695 0.928422
## CA30 CA31 CA32 CA33 CA34
## Eigenvalue 0.010939 0.010493 0.010011 0.009552 0.009149
## Proportion Explained 0.009587 0.009197 0.008775 0.008372 0.008018
## Cumulative Proportion 0.938009 0.947207 0.955981 0.964354 0.972372
## CA35 CA36 CA37 CA38
## Eigenvalue 0.008627 0.008251 0.007549 0.007094
## Proportion Explained 0.007561 0.007232 0.006617 0.006218
## Cumulative Proportion 0.979933 0.987166 0.993782 1.000000
arctic_cca_tjan <- vegan::cca(arctic_sqrt ~ arctic.env$tjan)
arctic_cca_tjul_sum <- summary(arctic_cca_tjul)
arctic_cca_tjul_sum$cont
## $importance
## Importance of components:
## CCA1 CA1 CA2 CA3 CA4 CA5
## Eigenvalue 0.1530 0.1547 0.09566 0.07269 0.05484 0.04373
## Proportion Explained 0.1341 0.1356 0.08384 0.06371 0.04806 0.03832
## Cumulative Proportion 0.1341 0.2697 0.35349 0.41720 0.46527 0.50359
## CA6 CA7 CA8 CA9 CA10 CA11
## Eigenvalue 0.03706 0.03312 0.03183 0.02889 0.02617 0.02573
## Proportion Explained 0.03248 0.02903 0.02790 0.02532 0.02293 0.02256
## Cumulative Proportion 0.53607 0.56510 0.59300 0.61832 0.64126 0.66381
## CA12 CA13 CA14 CA15 CA16 CA17
## Eigenvalue 0.02362 0.02275 0.02196 0.02068 0.02044 0.01879
## Proportion Explained 0.02071 0.01994 0.01925 0.01813 0.01792 0.01647
## Cumulative Proportion 0.68452 0.70446 0.72370 0.74183 0.75975 0.77622
## CA18 CA19 CA20 CA21 CA22 CA23
## Eigenvalue 0.01862 0.01791 0.01711 0.01633 0.01488 0.01449
## Proportion Explained 0.01632 0.01570 0.01500 0.01431 0.01304 0.01270
## Cumulative Proportion 0.79254 0.80823 0.82323 0.83754 0.85058 0.86329
## CA24 CA25 CA26 CA27 CA28 CA29
## Eigenvalue 0.01352 0.01317 0.01281 0.01239 0.011328 0.011098
## Proportion Explained 0.01185 0.01154 0.01123 0.01086 0.009929 0.009727
## Cumulative Proportion 0.87514 0.88668 0.89791 0.90877 0.918695 0.928422
## CA30 CA31 CA32 CA33 CA34
## Eigenvalue 0.010939 0.010493 0.010011 0.009552 0.009149
## Proportion Explained 0.009587 0.009197 0.008775 0.008372 0.008018
## Cumulative Proportion 0.938009 0.947207 0.955981 0.964354 0.972372
## CA35 CA36 CA37 CA38
## Eigenvalue 0.008627 0.008251 0.007549 0.007094
## Proportion Explained 0.007561 0.007232 0.006617 0.006218
## Cumulative Proportion 0.979933 0.987166 0.993782 1.000000
varpart(arctic_sqrt, arctic.env$tjul, arctic.env$tjan)
##
## Partition of variance in RDA
##
## Call: varpart(Y = arctic_sqrt, X = arctic.env$tjul,
## arctic.env$tjan)
##
## Explanatory tables:
## X1: arctic.env$tjul
## X2: arctic.env$tjan
##
## No. of explanatory tables: 2
## Total variation (SS): 22419
## Variance: 27.109
## No. of observations: 828
##
## Partition table:
## Df R.squared Adj.R.squared Testable
## [a+b] = X1 1 0.22897 0.22804 TRUE
## [b+c] = X2 1 0.03129 0.03012 TRUE
## [a+b+c] = X1+X2 2 0.27042 0.26865 TRUE
## Individual fractions
## [a] = X1|X2 1 0.23853 TRUE
## [b] 0 -0.01049 FALSE
## [c] = X2|X1 1 0.04061 TRUE
## [d] = Residuals 0.73135 FALSE
## ---
## Use function 'rda' to test significance of fractions of interest
The shared variance explained for July and January temperature is 26.8%Â The variance explained by July temp is 22.9 percent and by January temp is 3.1%.
arctic_cca_climate <- vegan::cca(arctic_sqrt ~ arctic.env$tjan + arctic.env$tjul + arctic.env$ptotal + arctic.env$sjul)
arctic_cca_clim_sum <- summary(arctic_cca_climate)
arctic_cca_clim_sum$cont
## $importance
## Importance of components:
## CCA1 CCA2 CCA3 CCA4 CA1 CA2
## Eigenvalue 0.1862 0.06987 0.02958 0.01459 0.08071 0.06992
## Proportion Explained 0.1632 0.06124 0.02593 0.01278 0.07074 0.06128
## Cumulative Proportion 0.1632 0.22446 0.25039 0.26317 0.33391 0.39520
## CA3 CA4 CA5 CA6 CA7 CA8
## Eigenvalue 0.06444 0.04416 0.03502 0.03424 0.03244 0.03084
## Proportion Explained 0.05648 0.03870 0.03069 0.03001 0.02843 0.02703
## Cumulative Proportion 0.45168 0.49038 0.52107 0.55108 0.57952 0.60655
## CA9 CA10 CA11 CA12 CA13 CA14
## Eigenvalue 0.02685 0.02568 0.02485 0.02264 0.02253 0.02110
## Proportion Explained 0.02354 0.02251 0.02178 0.01985 0.01974 0.01849
## Cumulative Proportion 0.63008 0.65259 0.67437 0.69422 0.71396 0.73246
## CA15 CA16 CA17 CA18 CA19 CA20
## Eigenvalue 0.01942 0.01872 0.01835 0.01775 0.01709 0.01666
## Proportion Explained 0.01702 0.01641 0.01608 0.01556 0.01498 0.01460
## Cumulative Proportion 0.74948 0.76589 0.78198 0.79753 0.81251 0.82711
## CA21 CA22 CA23 CA24 CA25 CA26
## Eigenvalue 0.01564 0.01446 0.01440 0.01330 0.01278 0.01265
## Proportion Explained 0.01371 0.01268 0.01262 0.01165 0.01120 0.01108
## Cumulative Proportion 0.84082 0.85349 0.86611 0.87776 0.88897 0.90005
## CA27 CA28 CA29 CA30 CA31 CA32
## Eigenvalue 0.01199 0.011282 0.010955 0.010515 0.010442 0.009709
## Proportion Explained 0.01051 0.009889 0.009602 0.009216 0.009152 0.008510
## Cumulative Proportion 0.91056 0.920444 0.930046 0.939262 0.948414 0.956924
## CA33 CA34 CA35 CA36 CA37
## Eigenvalue 0.009536 0.009113 0.008251 0.007829 0.007474
## Proportion Explained 0.008358 0.007988 0.007232 0.006862 0.006550
## Cumulative Proportion 0.965282 0.973270 0.980501 0.987363 0.993913
## CA38
## Eigenvalue 0.006945
## Proportion Explained 0.006087
## Cumulative Proportion 1.000000
January and July temperatures, annual precipitation and July sunshine explain 39.6% of the variance in the arctic pollen dataset.
# Creates triplot with temperature
plot(arctic_cca_climate)
# Adds ordination for each environmental variable
vegan::ordisurf(x = arctic_cca_climate, arctic.env$tjan)
##
## Family: gaussian
## Link function: identity
##
## Formula:
## y ~ s(x1, x2, k = 10, bs = "tp", fx = FALSE)
##
## Estimated degrees of freedom:
## 8.58 total = 9.58
##
## REML score: 2522.47
vegan::ordisurf(x = arctic_cca_climate, arctic.env$tjul)
##
## Family: gaussian
## Link function: identity
##
## Formula:
## y ~ s(x1, x2, k = 10, bs = "tp", fx = FALSE)
##
## Estimated degrees of freedom:
## 8.84 total = 9.84
##
## REML score: 1699.555
vegan::ordisurf(x = arctic_cca_climate, arctic.env$ptotal)
##
## Family: gaussian
## Link function: identity
##
## Formula:
## y ~ s(x1, x2, k = 10, bs = "tp", fx = FALSE)
##
## Estimated degrees of freedom:
## 8.5 total = 9.5
##
## REML score: 5803.674
vegan::ordisurf(x = arctic_cca_climate, arctic.env$sjul)
##
## Family: gaussian
## Link function: identity
##
## Formula:
## y ~ s(x1, x2, k = 10, bs = "tp", fx = FALSE)
##
## Estimated degrees of freedom:
## 8.55 total = 9.55
##
## REML score: 2395.364
All the variables are explaining different directions of variance in the data. The july temperature and total precipitation point in approximately the same direction. January and July temperature are also relatively close.
Question 3
# Subsets 4 species of interest from arctic pollen data
subset <- arctic.pollen[,c(13,7,25,26)]
# Estimate WA optima and tolerances for chosen taxa
subset_optima <- analogue::optima(subset, arctic.env$tjul)
subset_tolerance <- tolerance(subset, arctic.env$tjul)
# Plots abundances of taxa as function of July temp with unimodal response
plot(arctic.env$tjul, arctic.pollen[,c(13)], xlab = 'F.CYPE Abundance')
lines(arctic.env$tjul,(dnorm(arctic.env$tjul,subset_optima[1],subset_tolerance[1])*100))
plot(arctic.env$tjul, arctic.pollen[,c(7)], xlab = 'F.PPIN Abundance')
lines(arctic.env$tjul,(dnorm(arctic.env$tjul,subset_optima[2],subset_tolerance[2])*100))
plot(arctic.env$tjul, arctic.pollen[,c(25)], xlab = 'F.PPIN Abundance')
lines(arctic.env$tjul,(dnorm(arctic.env$tjul,subset_optima[3],subset_tolerance[3])*100))
plot(arctic.env$tjul, arctic.pollen[,c(26)], xlab = 'F.PPIN Abundance')
lines(arctic.env$tjul,(dnorm(arctic.env$tjul,subset_optima[4],subset_tolerance[4])*100))
# Runs WA and summarizes results
arctic_WA <- rioja::WA(arctic_sqrt, arctic.env$tjul)
arctic_WA_sum <- summary(arctic_WA)
##
## Method : Weighted Averaging
## Call : rioja::WA(y = arctic_sqrt, x = arctic.env$tjul)
##
## Tolerance DW : No
## Monotonic deshrink : No
## No. samples : 828
## No. species : 39
## Cross val. : none
##
## Deshrinking regression coefficients:
## Inverse d/s Classical d/s
## wa.b0 -29.6874 8.1298
## wa.b1 4.1067 0.1526
##
## Performance:
## RMSE R2 Avg.Bias Max.Bias Skill
## WA.inv 2.1638 0.6268 0 5.1059 62.6822
## WA.cla 2.7331 0.6268 0 3.3543 40.4652
##
##
## Fitted values
## WA.inv WA.cla
## 1 9.716231 9.599130
## 2 8.292863 7.328364
## 3 8.351207 7.421443
## . . .
## . . .
## 831 9.525978 9.295611
##
## Species coefficients
## Optima
## F.PABI 12.307671
## F.BALN 10.490190
## F.CAMB 9.389703
## . .
## . .
## F.ULMA 12.812432
# Estimates July temperature optima for all taxa in the arctic pollen dataset
arctic_optima <- analogue::optima(arctic.pollen, arctic.env$tjul)
plot(arctic_optima, arctic_cca_tjul$CCA$v)
The Arctic pollen optima line up quite well with the species scores calculated for the CCA.
# Predict July temperatures
arctic_WA_predict <- predict(arctic_WA)
# Compares predicted July temps to CCA site scores
plot(arctic_cca_tjul$CCA$u, arctic_WA_predict[,1])
There’s a reasonably strong correlation between the CCA site scores and the WA predicted July temperatures
# Compare apparent and cross-validated (crossval(my.wa.model)) predictions of July temperature. Also compare performance statistics (performance(my.cv.wa.model)).
arctic_WA_cv <- rioja::crossval(arctic_WA)
## Cross-validating:
##
|
| | 0%
|
| | 1%
|
|= | 1%
|
|= | 2%
|
|== | 2%
|
|== | 3%
|
|== | 4%
|
|=== | 4%
|
|=== | 5%
|
|==== | 5%
|
|==== | 6%
|
|==== | 7%
|
|===== | 7%
|
|===== | 8%
|
|====== | 9%
|
|====== | 10%
|
|======= | 10%
|
|======= | 11%
|
|======== | 12%
|
|======== | 13%
|
|========= | 13%
|
|========= | 14%
|
|========= | 15%
|
|========== | 15%
|
|========== | 16%
|
|=========== | 16%
|
|=========== | 17%
|
|=========== | 18%
|
|============ | 18%
|
|============ | 19%
|
|============= | 19%
|
|============= | 20%
|
|============= | 21%
|
|============== | 21%
|
|============== | 22%
|
|=============== | 22%
|
|=============== | 23%
|
|=============== | 24%
|
|================ | 24%
|
|================ | 25%
|
|================= | 25%
|
|================= | 26%
|
|================= | 27%
|
|================== | 27%
|
|================== | 28%
|
|=================== | 29%
|
|=================== | 30%
|
|==================== | 30%
|
|==================== | 31%
|
|==================== | 32%
|
|===================== | 32%
|
|===================== | 33%
|
|====================== | 33%
|
|====================== | 34%
|
|====================== | 35%
|
|======================= | 35%
|
|======================= | 36%
|
|======================== | 36%
|
|======================== | 37%
|
|======================== | 38%
|
|========================= | 38%
|
|========================= | 39%
|
|========================== | 39%
|
|========================== | 40%
|
|========================== | 41%
|
|=========================== | 41%
|
|=========================== | 42%
|
|============================ | 42%
|
|============================ | 43%
|
|============================ | 44%
|
|============================= | 44%
|
|============================= | 45%
|
|============================== | 45%
|
|============================== | 46%
|
|============================== | 47%
|
|=============================== | 47%
|
|=============================== | 48%
|
|================================ | 49%
|
|================================ | 50%
|
|================================= | 50%
|
|================================= | 51%
|
|================================== | 52%
|
|================================== | 53%
|
|=================================== | 53%
|
|=================================== | 54%
|
|=================================== | 55%
|
|==================================== | 55%
|
|==================================== | 56%
|
|===================================== | 56%
|
|===================================== | 57%
|
|===================================== | 58%
|
|====================================== | 58%
|
|====================================== | 59%
|
|======================================= | 59%
|
|======================================= | 60%
|
|======================================= | 61%
|
|======================================== | 61%
|
|======================================== | 62%
|
|========================================= | 62%
|
|========================================= | 63%
|
|========================================= | 64%
|
|========================================== | 64%
|
|========================================== | 65%
|
|=========================================== | 65%
|
|=========================================== | 66%
|
|=========================================== | 67%
|
|============================================ | 67%
|
|============================================ | 68%
|
|============================================= | 68%
|
|============================================= | 69%
|
|============================================= | 70%
|
|============================================== | 70%
|
|============================================== | 71%
|
|=============================================== | 72%
|
|=============================================== | 73%
|
|================================================ | 73%
|
|================================================ | 74%
|
|================================================ | 75%
|
|================================================= | 75%
|
|================================================= | 76%
|
|================================================== | 76%
|
|================================================== | 77%
|
|================================================== | 78%
|
|=================================================== | 78%
|
|=================================================== | 79%
|
|==================================================== | 79%
|
|==================================================== | 80%
|
|==================================================== | 81%
|
|===================================================== | 81%
|
|===================================================== | 82%
|
|====================================================== | 82%
|
|====================================================== | 83%
|
|====================================================== | 84%
|
|======================================================= | 84%
|
|======================================================= | 85%
|
|======================================================== | 85%
|
|======================================================== | 86%
|
|======================================================== | 87%
|
|========================================================= | 87%
|
|========================================================= | 88%
|
|========================================================== | 89%
|
|========================================================== | 90%
|
|=========================================================== | 90%
|
|=========================================================== | 91%
|
|============================================================ | 92%
|
|============================================================ | 93%
|
|============================================================= | 93%
|
|============================================================= | 94%
|
|============================================================= | 95%
|
|============================================================== | 95%
|
|============================================================== | 96%
|
|=============================================================== | 96%
|
|=============================================================== | 97%
|
|=============================================================== | 98%
|
|================================================================ | 98%
|
|================================================================ | 99%
|
|=================================================================| 99%
|
|=================================================================| 100%
plot(arctic_WA_cv$predicted[,1], arctic_WA_predict[,1])
arctic_WA_perf <- rioja::performance(arctic_WA)
arctic_WA_cv_perf <- rioja::performance(arctic_WA_cv)
arctic_WA_cv_perf
## $RMSE0
## [1] 3.54216
##
## $object
## RMSE R2 Avg.Bias Max.Bias Skill
## WA.inv 2.163844 0.6268224 -6.351707e-16 5.105933 62.68224
## WA.cla 2.733089 0.6268224 -5.405017e-15 3.354266 40.46518
##
## $crossval
## RMSE R2 Avg.Bias Max.Bias Skill
## WA.inv 2.179756 0.6213157 0.001413862 5.197317 62.13140
## WA.cla 2.745448 0.6219778 0.003391045 3.497582 39.92552
The performance statistics are overall very similar. The average bias for the cross-validated model is higher.
# Compare your MAT based reconstruction (week 8) to a WA based reconstruction of the same enironmental variable for the same site.
# Plot WA results
plot(arctic_WA)
title(main = "Arctic Pollen WA")
# Downloads modern pollen and climate data
data(Pollen)
data(Climate)
# Assigns climate variables to vectors
t_max <- Climate[,26]
# Cleans counts
Pollen_corrected <- replace(Pollen,is.na(Pollen),0)
# Calculates pollen percentages
Pollen_percentages <- 100*(Pollen_corrected/rowSums(Pollen_corrected))
# Runs MAT on modern pollen percentages and max temp
MAT.t_max <- rioja::MAT(Pollen_percentages, t_max, lean = FALSE)
## Warning in rioja::MAT(Pollen_percentages, t_max, lean = FALSE): Inter-sample distances < 1.0E-6 found (duplicate samples?
## These have been replaced by 1.0E-6
# Plots MAT results
plot(MAT.t_max)
title(main = "Arctic Pollen MAT")
# cross-validates the MAT using the leave-one-out technique
cv.tmax.mat.model <- rioja::crossval(MAT.t_max, cv.method='lgo', verbose=FALSE)
cv.tmax.mat.model
##
## Method : Modern Analogue Technique
## Call : MAT(y = y, x = x, dist.method = "sq.chord", k = 5, lean = FALSE)
##
## Distance : sq.chord
## No. samples : 4833
## No. species : 134
## Cross val. : lgo : no. groups = 10
##
##
## Performance:
## RMSE R2 Avg.Bias Max.Bias Skill
## N01 2.1141 0.8631 -0.0325 2.3925 85.8239
## N02 1.8764 0.8893 -0.0349 2.8825 88.8321
## N03 1.8253 0.8947 -0.0327 3.0750 89.4329
## N04 1.8245 0.8946 -0.0470 3.0513 89.4418
## N05 1.8325 0.8937 -0.0550 3.0790 89.3496
## N01.wm 2.1141 0.8631 -0.0325 2.3925 85.8239
## N02.wm 1.8580 0.8914 -0.0373 2.8148 89.0502
## N03.wm 1.7919 0.8985 -0.0340 3.0375 89.8155
## N04.wm 1.7825 0.8994 -0.0469 3.0139 89.9221
## N05.wm 1.7798 0.8997 -0.0554 3.0334 89.9534
## N01_XVal 2.1385 0.8596 -0.0562 2.3675 85.4949
## N02_XVal 1.9010 0.8864 -0.0318 2.9163 88.5376
## N03_XVal 1.8572 0.8911 -0.0487 2.9417 89.0595
## N04_XVal 1.8463 0.8921 -0.0479 2.9719 89.1883
## N05_XVal 1.8481 0.8918 -0.0567 2.8695 89.1676
## N01.wm_XVal 2.1385 0.8596 -0.0562 2.3675 85.4949
## N02.wm_XVal 1.8829 0.8885 -0.0379 2.8615 88.7548
## N03.wm_XVal 1.8226 0.8951 -0.0501 2.9130 89.4642
## N04.wm_XVal 1.8057 0.8968 -0.0508 2.9460 89.6580
## N05.wm_XVal 1.7956 0.8979 -0.0587 2.8476 89.7732
# Determines RMSE, performance statistics, and cross-validated predictions' performance
perf.cv.tmax.mat.model <- rioja::performance(cv.tmax.mat.model)
perf.cv.tmax.mat.model
## $RMSE0
## [1] 5.61502
##
## $object
## RMSE R2 Avg.Bias Max.Bias Skill
## N01 2.114118 0.8630960 -0.03252638 2.392500 85.82393
## N02 1.876448 0.8892978 -0.03485413 2.882500 88.83213
## N03 1.825280 0.8947197 -0.03265742 3.075000 89.43289
## N04 1.824513 0.8946198 -0.04704635 3.051250 89.44176
## N05 1.832459 0.8936511 -0.05501345 3.079000 89.34960
## N01.wm 2.114118 0.8630960 -0.03252638 2.392500 85.82393
## N02.wm 1.858038 0.8914389 -0.03734852 2.814839 89.05018
## N03.wm 1.791929 0.8984843 -0.03396247 3.037489 89.81551
## N04.wm 1.782531 0.8993890 -0.04687621 3.013941 89.92206
## N05.wm 1.779758 0.8996679 -0.05541570 3.033404 89.95339
##
## $crossval
## RMSE R2 Avg.Bias Max.Bias Skill
## N01 2.138512 0.8596398 -0.05617629 2.367500 85.49490
## N02 1.901032 0.8863952 -0.03183323 2.916250 88.53758
## N03 1.857250 0.8911101 -0.04865853 2.941667 89.05948
## N04 1.846283 0.8921158 -0.04793606 2.971875 89.18830
## N05 1.848052 0.8918281 -0.05665218 2.869500 89.16757
## N01.wm 2.138512 0.8596398 -0.05617629 2.367500 85.49490
## N02.wm 1.882936 0.8885027 -0.03792799 2.861488 88.75476
## N03.wm 1.822572 0.8950587 -0.05014030 2.913006 89.46421
## N04.wm 1.805730 0.8967768 -0.05080119 2.945995 89.65804
## N05.wm 1.795646 0.8978703 -0.05870765 2.847611 89.77322
The MAT model is much more tightly grouped along a one-to-one line than the WA.